List:General Discussion« Previous MessageNext Message »
From:James Rogers Date:July 14 1999 3:25am
Subject:Re: surviving the Slashdot effect
View as plain text  
At 03:50 PM 7/13/99 -0700, you wrote:
>>Oracle does have a higher overhead which makes it slower for read-heavy
>>apps, but the benefit is that Oracle can sustain extreme levels of
>>read-write concurrency without contention.  With Oracle, you can have, say,
>>50 different threads writing rows to a table simultaneously while another
>>200 threads read the same table, and all with out any significant
>>contention or lock waits.  Reads are never blocked by writes.  Anytime you
>>have a fair amount of writing and lots of users, Oracle will shine (as it
>>should since that is what it is designed to do).  The Oracle kernel is
>>designed to reduce contention to the minimum possible levels even under
>>heavy write loads; a user wanting to read a record will not notice a wait
>>or a significant performance loss due to a large number of threads writing
>>the same table or even the same row.  Even pathological cases are handled
>>gracefully.
>
>I certainly don't want to start a flamewar over something silly like this,
but
>I just cannot let this particular paragraph (and especially the last
sentence)
>go by without one comment:
>
>Not In My Experience.


I am by no means an Oracle (or any other database) zealot.  They all have
their strengths and weakness, and given a choice, I'll choose the one best
suited for the particular job I am doing.  If I could only choose one
database for everything, it would probably be DB2.  

I will respectfully submit that you have never worked on a truly well-tuned
Oracle implementation.

In my experience, 90% of Oracle application implementations are seriously
broken.  Poor performance is not the fault of Oracle if the developers and
DBAs are inadequately skilled at using it.  For a short while I worked as
an Oracle7 performance consultant (good money, but really boring).  When I
walked into a company, I could almost always guarantee 3-10x performance
improvement, simply because I was intimately familiar with behaviors of the
Oracle engine.  I have *never* seen an implementation of an Oracle
application that I could not make significant performance improvements to.
I have met a lot of very experienced Oracle people at very big companies
who weren't aware of performance issues that are fundamental to the smooth
functioning of the db.  And it isn't just up to the DBAs.  Huge performance
gains can be made by making intelligent decisions and smart code tweaks at
the application level.  Truly expert Oracle developers are far and few
between indeed.

YMMV <shrug>,

-James Rogers 
 jamesr@stripped
Thread
surviving the Slashdot effectThomas D. Wilkinson12 Jul
  • Re: surviving the Slashdot effectMichael R. Gile12 Jul
  • Re: surviving the Slashdot effectSrfRoG12 Jul
    • Re: surviving the Slashdot effectAdam Powell12 Jul
  • Re: surviving the Slashdot effectOrlando Andico12 Jul
  • Re: surviving the Slashdot effectSasha Pachev13 Jul
    • Re: surviving the Slashdot effectJames Rogers13 Jul
  • Re: surviving the Slashdot effectPatrick Greenwell13 Jul
    • Re: surviving the Slashdot effectThomas D. Wilkinson13 Jul
      • Re: surviving the Slashdot effectPatrick Greenwell13 Jul
        • Re: surviving the Slashdot effectThomas D. Wilkinson14 Jul
          • Re: surviving the Slashdot effectMichael Widenius30 Jul
        • Re: surviving the Slashdot effectTõnu Samuel14 Jul
Re: surviving the Slashdot effectDerick H Siddoway14 Jul
Re: surviving the Slashdot effectJames Rogers14 Jul
  • Re: surviving the Slashdot effectThomas D. Wilkinson14 Jul
    • Re: surviving the Slashdot effectJames Rogers17 Jul
Re: surviving the Slashdot effectDerick H Siddoway15 Jul
Re: surviving the Slashdot effectSasha Pachev15 Jul
  • Re: surviving the Slashdot effectDarrell Shifflett15 Jul
  • Re: surviving the Slashdot effectJames Rogers17 Jul
Re: surviving the Slashdot effectJames Rogers17 Jul
  • Re: surviving the Slashdot effectThomas D. Wilkinson17 Jul
    • Re: surviving the Slashdot effect (headed slightly off topic)\"Terrence W. Zellers\"18 Jul
Re: surviving the Slashdot effectAntony T Curtis19 Jul
  • Re: surviving the Slashdot effectSasha Pachev19 Jul
Re: surviving the Slashdot effectDerick H Siddoway19 Jul